AI.

Where IDU uses AI — and where it deliberately doesn’t.

4 questions

Does IDU use a large language model to generate forecasts?
No. Predictive Forecast Assumptions use classical time-series techniques trained on your own historical data. LLMs power the Help Agent and Insight Agent for narrative and navigation — not for producing the numbers. Finance-critical figures never depend on generative models.
Does our data train shared or external models?
No. Predictions are computed on your data, inside your tenant. Nothing is pooled, nothing is used to improve a shared model. The customer data boundary is hard and non-negotiable.
How are AI suggestions tracked for audit?
Every suggested value, every human override, and every final committed figure is timestamped and preserved. Auditors see exactly what the model produced, what the budget owner accepted or changed, and why.
Can we turn AI features off?
Yes — at the organisation, team, or user level. AI-assisted features are opt-in per module and per role. Organisations that run without AI-informed forecasting get the same governed planning environment without any prediction layer involved.